Joule Heist search

topics · updated 2026-09-25

Adsorption Cooling — Cold From Heat

confidence: high volatility: cold verified: 2026-09-25fresh

Adsorption cooling produces cold from low-grade heat (typically 60–95 °C, feasible from ~50 °C) using a solid porous adsorbent that alternately adsorbs and desorbs a refrigerant vapor instead of using an electric compressor. It is the keystone heat-driven cooling technology for waste-heat- and solar-thermal-rich sites, with silica-gel/water as the workhorse working pair. Strengths are benign refrigerants, few moving parts, and a very high electrical COP; limits are a low thermal COP (~0.3–0.7 for practical two-bed machines), intermittent operation, high-vacuum operation, and poor heat and mass transfer in the bed — the three structural problems the whole research program is organized around.

Adsorption cooling makes cold from heat instead of shaft work. Where a vapor-compression chiller uses an electric compressor, an adsorption chiller uses a solid porous adsorbent (silica gel, zeolite, activated carbon, or advanced MOF/composite materials) that alternately adsorbs and desorbs a refrigerant vapor — usually water — as it is cooled and heated. The driving energy is low-grade heat: 60–95 °C is the normal band and regeneration works from 50–60 °C, so it is the natural cooling technology wherever waste heat or solar-thermal heat already exists. It is the largest theme in this KB (113 of 150 registry items) because it is the keystone of a heat-driven cooling strategy.

It is one branch of sorption refrigeration — the other being liquid-sorbent absorption — and the stakes are scale: ~15 % of world electricity goes to refrigeration and air-conditioning, and A/C is ~45 % of building energy in hot regions (Wolak 2017).

Why it’s the keystone for the 601 Delaware roastery: a coffee roaster rejects large amounts of hot exhaust. An adsorption chiller turns that waste heat into chilled water for the building — cooling paid for by heat that was otherwise dumped. See 601 Cooling Strategy.

How it works (the cycle)

Every review in the set (Demir 2008, Elsheniti 2018, Kumar 2020) draws the same four legs on a Clapeyron diagram (ln P vs −1/T) along the adsorbent’s isosteres:

  1. Isosteric pre-heating — the saturated bed is isolated and warmed; pressure rises at constant loading to condenser pressure.
  2. Isobaric desorption / condensation (driven by heat) — hot water, exhaust or a solar collector keeps heating the bed; released vapor condenses and its heat is rejected to a cooling tower or ambient loop.
  3. Isosteric pre-cooling — the regenerated bed is isolated and cooled to evaporator pressure.
  4. Isobaric adsorption / evaporation (produces cold) — the cool, dry bed pulls vapor off the evaporator, whose evaporating refrigerant takes latent heat from the chilled-water loop: the cooling effect. The heat of adsorption is carried away by cooling water.

Thermal COP = evaporator cooling ÷ heat supplied in steps 1–2; specific cooling power (SCP) = evaporator cooling ÷ (adsorbent mass × cycle time). COP says how efficiently heat is used, SCP how big the machine is per ton. A single bed cools only in step 4, so Wolak’s classification splits intermittent single-adsorber machines (the basic solar-refrigerator layout, sometimes COP < 0.1 because the whole bed swings through its full temperature range each cycle) from continuous machines with two or more beds out of phase.

Advanced cycles

The bed’s sensible heat and heat of adsorption are the main losses, so cycle-level gains come from recovering heat between beds: heat recovery (routing the stream leaving a bed being cooled through the bed being heated) gives up to ~25 % COP — Elsheniti and Wolak agree, Demir cites up to ~50 % for uniform-temperature two-bed recovery; mass recovery (briefly connecting the high- and low-pressure beds at switch-over) helps most at low driving temperature, and combined heat + mass recovery adds > 10 % more (Elsheniti); thermal-wave cycles move a temperature front through the beds so ~80 % of desorption heat comes from adsorption heat (Critoph’s convective version: theoretical COP 0.9; a zeolite–water two-bed cooler ~1.0); cascade cycles use a zeolite–water high stage to drive a carbon–methanol low stage and reached an experimental COP of 1.06 (Douss & Meunier, via Demir); and cycle-time optimisation — an optimum exists per condition (Kumar’s survey: ~720 s cycles with 80 s mass- and 20 s heat-recovery phases) — so adaptive half-cycle timing beats fixed timing.

The working pairs (the heart of the machine)

Performance is set by the adsorbent + refrigerant pair — isotherm, kinetics, heat of adsorption, regeneration temperature, stability. Demir’s selection checklist: high refrigerant latent heat; non-corrosive, non-toxic; chemically and thermally stable; high capacity and conductivity; low cost; and a physical, reversible interaction. Consolidating Demir’s comparison table, Elsheniti’s material survey and Kumar’s review:

Working pairRegeneration tempUptake / heat of adsorptionNotes
Silica gel / watertypically 85 °C; works from 50–60 °C; never above ~120 °C (gel destroyed)0.20–0.40 kg/kg equilibrium but a usable swing ≤ 0.1 kg/kg; ~2500–2800 kJ/kgThe workhorse: cheap, benign, most lab and commercial chillers. Low swing → bulky.
Zeolite / water (13X, 4A, natural)70–250 °C band (Elsheniti); Kumar: ~200 °C in practice; Demir’s table: 30–350 °C13X 0.30, 4A 0.22, natural 0.11–0.12 kg/kg; 3300–4400 kJ/kgHigh uptake, stable, exhaust-heatable; but high heat of adsorption lowers COP and classic zeolites want heat too hot for low-grade sources. Aluminophosphates (AQSOA, EMM-8) are the exception — [[references/zeolite
Activated carbon / methanol (ethanol)~100 °C, < 120 °C (methanol → dimethyl ether above ~150 °C)0.32–0.45 kg/kg (fibres 0.55); 1400–2000 kJ/kgLow heat of adsorption helps COP; sub-atmospheric; fibres conduct badly (~0.09 W/m·K). The solar ice-maker pair.
Activated carbon / ammonia, CO₂, R-134aNH₃ suits ≥ 200 °C sourcesNH₃: 1800–2000 kJ/kgPressurised pairs (NH₃ ~1600 kPa at 40 °C condensing): no vacuum, but pressure vessels and (for NH₃) toxicity. Carbon/CO₂ sub-/transcritical cycles suit sites where water’s freezing and vacuum are unwanted — [[topics/carbon-dioxide-r744
Metal chlorides / ammonia40–80 °Capproaching 1 kg/kg (1 mol CaCl₂ binds 8 mol NH₃)Chemisorption: huge capacity, but swelling/agglomeration destabilise bare salts — hence salt-in-porous-matrix composites (SWS-1L: cooling COP 0.17–0.48 in a coated finned bed, Demir) — [[references/composite-salt-sorbents
MOFs / water or ethanolAl-fumarate from 60 °CMIL-101(Cr) ~1 g/g, MIL-101@GO 1.58 g/g; HKUST-1 +93 % over silica gelCPO-27(Ni): SCP 440 W/kg vs 310 for SAPO-34 at 130 °C. The frontier; cost, durability, scale-up open — [[references/adsorption-limitations-and-mitigations

For the hard numbers — COP/SCP by pair, the silica-gel benchmark, EMM-8 (COP 0.85 at 63 °C) and MFM-300 MOFs, bed-enhancement gains, hot-day derating, economics and desalination cogeneration — see Performance & Numbers.

The three structural problems

Demir’s 2008 review organised the field around three problems, and the framing still holds: (1) intermittent cooling — fixed by multiple beds, buffering and the recovery cycles above; (2) high-vacuum operation — water systems run far below atmospheric pressure, and Kumar’s rig identified vacuum maintenance as the reliability limitation (pressurised refrigerants trade this for pressure-vessel and toxicity issues); (3) poor heat and mass transfer in the bed — adsorbents are insulators, so the answer is bed design: uncoated beds (pellets, granules or fibres between fins, with voids for vapor transport) versus coated beds (adsorbent bonded onto tubes, fins or metal foam, including hydrothermally grown zeolite layers), plus conductivity additives (aluminium foams, 3-D graphene binders for zeolite 13X, metal and adhesive additives), condenser-embedded adsorbers and adsorber/evaporator geometry — the subject of Adsorbent Bed Engineering.

What it’s good for (and not)

Strengths. It runs on low-grade heat — waste heat, solar, geothermal, engine exhaust: Kumar’s single-bed silica-gel rig regenerated at 60 °C and reached COP 0.51 there, 0.66 at 70 °C and 0.67 at 80 °C (evaporator bottoming at 16.5 / 14.3 / 13.2 °C). It has few or no moving parts, is near-silent, and draws almost no electricity — Elsheniti’s SorTech prototype used 9 W for 7.5 kW of cooling (system COP 0.55), and the electrical COP (cooling per unit of electricity, the number that matters when heat is free) can reach ~10 against 3–5 for vapor compression. Refrigerants are benign and low-pressure (water, alcohols); the bed provides built-in thermal storage that suits intermittent solar drive (Demir); and the same cycle can cogenerate chilled and distilled water — a major sub-thread for arid regions.

Limits. Thermal COP is low — Elsheniti calls ~0.6 “typical”; practical two-bed machines land in ~0.3–0.7, basic intermittent units below 0.1, and only cascade/thermal-wave lab cycles exceed 1. Low COP plus low SCP means a large, heavy machine per ton and high first cost — the target of the registry’s cost of manufacturing adsorption chillers study, the $2.54 M DOE project and ARPA-E’s high-efficiency chiller program. It is intermittent by nature; vacuum tightness and weather dependence are the reliability caveats (Kumar); it needs a good heat-rejection loop because performance is very sensitive to condenser temperature; and salt-based systems trade capacity for corrosion-limited life (Demir).

Adsorption wins when the heat is free, and loses on first cost and footprint when it isn’t.

Solar-driven adsorption cooling

A large fraction of the collection is solar adsorption cooling — flat-plate, evacuated-tube or concentrating collectors supplying desorption heat, mostly silica-gel/water, from vaccine refrigerators to building-scale fields. The appeal is the coincidence of load and supply: peak sun ≈ peak cooling demand. Layouts range from single-bed units in which the collector is the adsorber to two-bed chillers with storage. Documented installations: Freiburg University Hospital (1999; 70 kW on 170 m² of evacuated tubes) and the Shanghai Green Building (15 kW SJTU chiller, 150 m² of collectors, system COP 0.35, solar COP 0.15, 85 °C regeneration; payback 5–8 years, ~2 if the hot water also serves showers) (Elsheniti, Wolak). Registry economics studies (Jordan, Perth) favour high-DNI, high-load climates — the San Antonio profile. See Solar Adsorption Cooling.

Commercial reality

Wolak’s commercialisation review lists silica-gel/water chillers from Nishiyodo, Mayekawa (zeolite–water above 70 kW), Macom and SorTech, and water–zeolite machines from InvenSor; the registry adds Fahrenheit (SorTech’s successor, rack-integrated units), ART-KLIMA, Emissionless, BryChill, OxiCool and Green Adsorption Chiller, plus data-center deployments. Waste-heat units have run on fishing boats, cars, a zeolite-13X/water locomotive and fuel-cell vehicles. The cautionary tale is also recorded: SunMark, BLM, Zeopower, EG Solar and Solaref all failed to commercialise intermittent zeolite–water units — the layout simplest in the lab has been hardest to sell. CoolSkin, an adsorption-cooling façade, is the building-integration endpoint (Radiative & Façade Cooling). Current vendors and deployments: Commercial Adsorption Chillers.

Bench-scale calibration. Wolak’s own rig — ~170 g of coconut-shell activated carbon, 100 cm³ of methanol, band heaters, a fan and −0.9 bar of vacuum — desorbed for an hour at 120 °C and dropped the condenser side 10.4 °C in one cycle from 25 °C (4.9 and 4.7 °C over a double cycle from 28 °C). That is the whole machine in miniature, and why Building a DIY Adsorption Chiller exists.

What’s new (2025–2026)

An August 2026 research round found five moving fronts; provenance lives in the linked cluster articles (their sources are not in this article’s set): (1) an architecture generation shift — the Frazzica/Mittelbach “third-generation” adsorption heat pump (direct evaporation/condensation at the adsorber, liquid-phase refrigerant distribution, no large vacuum valves), fielded at a 200 kW brewery and an 80 kW pharma site via EU RE-WITCH (Commercial Adsorption Chillers); (2) the driving-temperature floor dropping toward data-center coolant (a 50 °C-driven encapsulated-salt chiller; Exora’s ~35 °C MOF claim) while hyperscale waste-heat cooling is being won by absorption and district-heat export; (3) MOF benchmarks jumped (cascaded COP 1.63, MOF-303 models near 1000 W·kg⁻¹) but durability didn’t (Performance & Numbers, Limitations & Mitigations); (4) buildable working-pair news — carbon/ethanol solar ice makers, zeolite 13X–water ranked the most practical long-life pair (Solar Adsorption Cooling); (5) materials strategy pivoted to salt confinement (Composite Salt Sorbents), and the atmospheric-water-harvesting field keeps confirming that bed-scale mass transfer, not sorbent g/g records, limits devices — Demir’s third problem restated (Adsorbent Bed Engineering).

See also

Sources